How ZoomInfo Scaled Real-Time Recommendations with Pinecone

​ZoomInfo's Applied AI team set out to change how users find the right people to contact. Instead of asking users to search, filter, and sort manually, ZoomInfo wanted to recommend the most relevant contacts the moment someone opens a company profile.

​At production scale, that meant serving thousands of recommendation requests per second and returning results end-to-end in under a second across ZoomInfo's entire customer base. With Pinecone, ZoomInfo scaled the system to serve 50x more peak customer requests, improve relevancy and recall by 2x, and increase user engagement by 50%.

​Join Keith Corbalis, Engineering Manager at Pinecone, and Tamiro Scholer, Senior Data Scientist at ZoomInfo, for a technical walkthrough of how ZoomInfo built and scaled its real-time recommendation system. They'll cover how the system evolved from proof of concept to production, the scaling limits the team hit along the way, and why ZoomInfo chose Pinecone Dedicated Read Nodes (DRN) to support their high-throughput recommendation workload.

​You'll leave with:

• ​A real account of taking a recommendation system from proof of concept to production

• ​Practical lessons for scaling real-time recommendations across large user bases

• ​A technical look at how Pinecone DRN supports low-latency recommendations at scale

​When: Wednesday, June 24, 9:00 to 10:00 AM PT / 12:00 to 1:00 PM ET